Homework 0

Due 2017/09/28

This homework consists of two parts. The first part you may write up with any tool you choose, but must be submitted as a .pdf file.

For the second part, the only part that requires a submission is 2(d), but you should complete 2(a)-(c) beforehand. Submit the R script (.R) you generate.

If you already know and use Rmarkdown, please feel free to use it here. Submit the Rmarkdown file (.Rmd), as well as a “Knit” version as a .pdf.

Submit your files on canvas.

  1. Your task is to find a news article reporting the results of a scientific study and also find the original source (usually a journal article) for the study, then complete the following tasks (a)-(e).

    Try to choose a study that, at least from the news article, appears to be either a random sampling study or randomized experiment (or has elements of both, but avoid studies that appear to meta-analyses, economic analyses, or don’t involve data).

    Finding the original source can be tricky, a good news article will contain a link to the study (this is good reporting, but can be rare). If there is no link you’ll need to look for clues like the author’s name and affiliation or the journal name and do some research, which should lead you to a title for the article. Then try a google scholar search to see if you can turn up the full article. It’s also worth trying a search on the library website for articles that appear to be behind a paywall, but OSU may have bought access to.

    Don’t underestimate how long this can take, you may go through a many articles before finding one that is appropriate and for which you can find the full original source.

    1. Report the headline of the news article, and provide a link to the news article.
    2. Does the headline imply population or causal inference, or both, or neither? Make sure you justify your answer by describing what population inferences are being made to, and/or what the treatment/intervention is.
    3. Report a full citation for the original source.
    4. What inferences are statistically justified? Justify your answer by quoting relevant parts of the the original source.
    5. Rewrite the news article headline to more faithfully represent the inferences supported by the study.
  2. This part requries you to get up and running with R and RStudio. If you are new to R, I’ll repeat my advice from class:
    1. Work through sections 3.1 and 3.2 of R for Data Science. What does work through mean? It means you read the material, but you also have RStudio open and are running code as it comes up in the Chapter. You might even try altering that code to test your understanding.

    2. Work through Chapter 4

    3. Work through Chapter 6

    4. Now return to 3.2.4. Create an R script that includes the answers to all the exercises in 3.2.4. Your script should include all code that you run to help you answer those questions, and the answers as code comments (code comments are lines in the script that start with #).